Identification of Constitutive Phosphorylation Sites on the Epstein-Barr Virus ZEBRA Protein*

ZEBRA, the product of the Epstein-Barr virus gene bzlf1, and a member of the AP-1 subfamily of basic zipper (bZIP) transcription factors, is necessary and sufficient to disrupt viral latency and to initiate the viral lytic cycle. Two serine residues of ZEBRA, Ser167 and Ser173, are substrates for casein kinase 2 (CK2) and are constitutively phosphorylated in vivo. Phosphorylation of ZEBRA at its CK2 sites is required for proper temporal regulation of viral gene expression. Phosphopeptide analysis indicated that ZEBRA contains additional constitutive phosphorylation sites. Here we employed a co-migration strategy to map these sites in vivo. The cornerstone of this strategy was to correlate the migration of 32P- and 35S-labeled tryptic peptides of ZEBRA. The identity of the peptides was revealed by mutagenesis of methionine and cysteine residues present in each peptide. Phosphorylation sites within the peptide were identified by mutagenesis of serines and threonines. ZEBRA was shown to be phosphorylated at serine and threonine residues, but not tyrosine. Two previously unrecognized phosphorylation sites of ZEBRA were identified in the NH2-terminal region of the transactivation domain: a cluster of weak phosphorylation sites at Ser6, Thr7, and Ser8 and a strong phosphorylation site at Thr14. Thr14 was embedded in a MAP kinase consensus sequence and could be phosphorylated in vitro by JNK, despite the absence of a canonical JNK docking site. Thus ZEBRA is now known to be constitutively phosphorylated at three distinct sites.

Phosphorylation, the most common post-translational modification of proteins, modulates the transcriptional potential of transcription factors by several discrete mechanisms. For example, phosphorylation of the NF-B subunit p65 at Ser 276 , c-Jun at Ser 63 -Ser 73 , and cAMP-response element-binding protein at Ser 133 , increases the transcriptional activity of these proteins by increasing their affinity for the transcriptional co-activator, CBP 2 (1)(2)(3). However, phosphorylation of c-Myb, Max, and c-Jun by CK2 or phosphorylation of OCT1 by protein kinase A at sites nearby or in the DNA binding domain of these activators negatively impacts their DNA binding activities (4 -6). By contrast, phosphorylation of p53 at its COOH terminus enhances the DNA binding activity of the protein (7)(8)(9). In addition, phosphorylation plays an important role in nuclear translocation of several transcription factors either by modifying the transcription factor itself, e.g. signal transducers and activators of transcription, or by modifying an inhibitory protein that sequesters the transcription factor in the cytoplasm, e.g. IB (10,11).
ZEBRA, a transcription activator encoded by the Epstein-Barr virus gene bzlf1 and a member of the basic zipper (bZIP) family of proteins can be considered to be composed of five functional regions: a transactivation domain (aa 1-167), a regulatory domain (aa 168 -177), a basic DNA binding domain (aa 178 -194), a coiled-coil dimerization domain (aa 195-227), and an accessory activation domain (aa 228 -245) (12)(13)(14)(15)(16)(17)(18)(19). Fourteen of 16 amino acids in the DNA binding domain of ZEBRA fit the DNA recognition consensus sequence present in other bZIP family members. This sequence is BB-BN-AA-B-R-BB, where B is any basic residue, N is asparagine, A is alanine, and R is arginine (20). The two exceptions are positions Ser 186 in ZEBRA, which is an alanine in the consensus and position Phe 193 , which is basic in the consensus. Even though all bZIP proteins share this similar DNA binding motif, each protein recognizes different response elements. ZEBRA binds to an array of sites known as ZEBRA response elements; in addition, the protein binds to the AP-1 consensus sequence, known as a TPA response element (21,22).
ZEBRA is a multifunctional protein. Its capacity to disrupt viral latency resides partially in its function as a transcriptional activator and partially in its role as a DNA replication protein (23)(24)(25). ZEBRA activates the promoter of brlf1, a gene expressing a second EBV encoded lytic cycle transcription factor, Rta (24, 26 -28). The two proteins function individually or synergistically to activate a cascade of lytic cycle genes essential for viral replication and virion production (29,30). ZEBRA also serves as a repressor of Rta-activated late genes during the early phase of the lytic cycle; thus ZEBRA temporally inhibits late gene expression prior to viral replication (29,31). As an essential factor required for viral DNA replication, the ZEBRA protein mediates interaction among several virally encoded protein components of the viral lytic replication machinery and binds to DNA encompassing the viral origin of lytic replication (32)(33)(34)(35)(36). Presumably, ZEBRA escorts these proteins to the site of replication and stabilizes the whole complex (36). Recently, the importance of ZEBRA binding to the viral lytic origin of replication in the process of viral DNA amplification was demonstrated by showing that a mutant Epstein-Barr virus, in which the ZEBRA binding sites in viral origin of lytic replication were exchanged for papilloma E2-binding sites, was markedly impaired in replication (37).
Despite the critical role of phosphorylation in modulating the activity of several members of the bZIP family of transcription factors, information about the role of phosphorylation in the function of ZEBRA is limited. ZEBRA, a phosphoprotein, is constitutively phosphorylated at multiple sites (38). Previous studies have demonstrated that the phosphorylation state of ZEBRA changes following treatment of EBV-infected cells with phorbol ester, a protein kinase C agonist that is capable of inducing the EBV lytic cycle (38,39). Nonetheless, ectopic expression of ZEBRA in the absence of such exogenous inducing agents is sufficient to activate the whole lytic cycle of EBV (40). Thus constitutive, but not inducible, phosphorylation of ZEBRA is likely to be adequate to establish its function as a lytic cycle switch. Moreover, the phosphorylation state of ZEBRA is not detectably altered in response to induction of the viral lytic cycle following transfection of ZEBRA expression vectors into cells harboring latent EBV. The phosphopeptide pattern of wild type ZEBRA and a mutant, Z(S186A), which is incompetent to activate the lytic cycle, are identical (40). Recently, we mapped two constitutive phosphorylation sites in vivo, Ser 167 and Ser 173 . Phosphorylation of ZEBRA at these two residues was found to be essential for the ability of the protein to act as a repressor of Rta-mediated activation of a late protein, blrf2 during early times in the lytic cycle (31). In this report, we further examine the constitutive phosphorylation of ZEBRA in vivo and identify several new phosphorylation sites. The strategies we utilized to analyze in vivo phosphorylation of ZEBRA are generally applicable to mapping the phosphorylation sites of proteins expressed at low levels in eukaryotic cells.

EXPERIMENTAL PROCEDURES
Cells and Transient Transfection-The human Burkitt lymphomaderived cell line HH514-16, an EBV-positive cell line permissive for viral replication, was maintained in RPMI 1640 medium supplemented with 8% fetal calf serum, penicillin, streptomycin, and amphotericin B (41). The HKB5/Cl8 cell line (HKB: hybrid kidney B-cell), a kind gift from Dr. Myung-Sam Cho, was generated by fusing HH514-16 cells with the 293 human embryonic kidney cell line. HKB5/Cl8 cells carry the EBV genome of HH514-16 cells. The cells were grown at 37°C in a humidified atmosphere containing 5% CO 2 . Transient transfection of HH514-16 cells was performed by means of electroporation. A mixture of 10 7 cells and 10 g of plasmid DNA were subjected to 960 microfarads and 250 volts using a Bio-Rad Gene Pulser.
Expression Vectors-Constitutive expression of ZEBRA from both pHD1013-gZ and FLAG-gZ plasmids was driven by the cytomegalovirus immediate early promoter. An EBV genomic DNA fragment (nucleotides 102115 to 103181) containing the bzlf1 gene was cloned into the eukaryotic expression vector, pHD1013, using the NaeI and NcoI restriction sites (42). To generate FLAG-gZ, two HindIII sites in the bzlf1 open reading frame were destroyed by site-directed mutagenesis (QuikChange, Stratagene). Neither change resulted in any amino acid alterations. The HindIII-resistant sequences were inserted into pHM201-FLAG (a gift from M. Hochstrasser, Yale University) as a HindIII-NotI fragment at compatible sites. The bacterial expression vector, pET-22b/cDNA ZEBRA, has been previously described (40). All point mutations were generated by the QuikChange site-directed mutagenesis system (Stratagene).
Bacterial Expression and Purification of ZEBRA-Escherichia coli (BL21) cells were transformed with pET22b/cDNA-Z or pET22b/ cDNA-Z(T14A). ZEBRA expression was induced by growing the cells in Luria-Bertani (LB) medium containing 1 mM isopropyl ␤-D-thiogalactopyranoside for 2 h at 37°C. The induced culture was centrifuged at 5000 ϫ g for 10 min at 4°C. The pellet was re-suspended in nickel column binding buffer containing 5 mM imidazole, 0.5 M NaCl, and 20 mM Tris-HCl (pH 7.9). ZEBRA was extracted by sonicating the cell lysate in binding buffer containing 6 M urea. The cell extract was cleared by centrifugation at 20,000 ϫ g for 15 min at 4°C, and the supernatant was loaded on a nickel affinity column. The retained ZEBRA protein was eluted by a stepwise gradient of imidazole ranging from 0 to 1 M dissolved in 10 mM Tris-HCl (pH 7.9), 0.25 M NaCl, and 6 M urea. Column fractions containing the ZEBRA protein were pooled; the purity of ZEBRA was examined by silver nitrate staining of an SDS-polyacrylamide gel. Purified ZEBRA proteins were refolded by dialysis against gradually decreasing concentrations of urea.
In Vitro Phosphorylation of ZEBRA by c-Jun NH 2 -terminal Kinase (JNK)-JNK phosphorylation of ZEBRA was carried out by mixing 500 ng of purified recombinant ZEBRA protein with JNK assay buffer (20 mM MOPS, pH 7.2, 25 mM ␤-glycerol phosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol) in the presence of 1 Ci of [␥-32 P]ATP, 7.5 mM MgCl 2 , and 50 M cold ATP. The phosphorylation reaction was initiated by adding different concentrations of recombinant JNK2 (Upstate) (see Fig. 6). After 30 min of incubation at 30°C, the reaction was stopped by adding SDS-PAGE sample buffer. The level of phosphorylation for each protein was assessed by SDS-PAGE. The gel was first stained with silver nitrate (Bio-Rad), then dried and autoradiographed.
Metabolic Labeling and Immunoprecipitation-Twelve hours after transfection of 1.5 ϫ 10 7 HH514-16 cells with 15 g of plasmid DNA, the cells were incubated in 3 ml of phosphate or methionine and cysteine-free RPMI 1640 medium (ICN) containing 1.6 mCi of [ 32 P i ]orthophosphate or 0.45 mCi of [ 35 S]methionine (70%) and [ 35 S]cysteine (15%) (Tran 35 S-label, MP Biomedicals), respectively. Labeling was carried out for 6 h at 37°C in a CO 2 incubator. Cell lysates were prepared by re-suspending the cells in 200 l of lysis buffer (50 mM Tris-HCl, pH 7.5, 0.25 M NaCl, 0.1% SDS, 0.1% Triton X-100, 5 mM EDTA, 50 mM NaF, 0.1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, 2 g of aprotinin per ml). Immunoprecipitations were performed as described previously (40). Briefly, ZEBRA was immunoprecipitated with a rabbit polyclonal antibody against the NH 2 -terminal domain of ZEBRA (43). The ZEBRA-antibody complex was captured with a mixture of protein A/protein G-coated agarose beads (Invitrogen). The immunoprecipitated proteins were resolved using SDS-8% PAGE; the gel was dried and autoradiographed.
Peptide Mapping and Phosphoamino Acid Analysis-Peptide mapping and phosphoamino acid analysis were carried out as described in the Hunter thin layer peptide mapping electrophoresis system manual. The radiolabeled ZEBRA band was excised from a dried SDS-PAGE gel and the protein was extracted in 50 mM ammonium bicarbonate, 10% ␤-mercaptoethanol, and 0.2% SDS. ZEBRA was separated from the gel slurry by centrifugation. The protein was precipitated from the supernatant with 20% trichloroacetic acid in the presence of 20 g of bovine albumin. The pellet was washed twice with ethanol and oxidized in 50 l of performic acid at 0°C for 1 h. After oxidation the sample was diluted with water, frozen, and lyophilized. The pellet was resuspended in 50 mM ammonium bicarbonate and digested with 20 g of trypsin (Promega, sequencing grade) overnight at 37°C. The tryptic digest was resolved by electrophoresis on a thin layer cellulose plate (EM Science) at 1000 volts for 25 min in pH 1.9 buffer (2.5% formic acid and 7.8% acetic acid (v/v)). After the plate was dried at room temperature, the peptides were subjected to ascending chromatography in phosphochromatography buffer (32.5% n-butanol, 25% pyridine, and 7.5% acetic acid (v/v)). The labeled peptides were visualized by autoradiography at Ϫ70°C.
Phosphoamino acid analysis was performed by hydrolyzing 32 P-labeled ZEBRA in 6 N HCl for 60 min at 110°C. The solution was dried in a SpeedVac and reconstituted in pH 1.9 buffer. Cold phosphoamino acid standards (Sigma) were added to the sample and the mixture was subjected to thin layer electrophoresis in two dimensions, the first in pH 1.9 buffer and the second in pH 3.5 buffer (5% acetic acid and 0.5% pyridine (v/v)). The plate was dried, sprayed with 5% ninhydrin (Sigma) dissolved in methanol, and incubated at 65°C for 5 min. Radiolabeled amino acids were visualized by autoradiography and their position was compared with the unlabeled reference phosphoamino acids.

ZEBRA Is Phosphorylated in Vivo at Serine and Threonine Residues-
As a prelude to mapping the phosphorylation sites on ZEBRA we performed phosphoamino acid analysis of the protein. ZEBRA was expressed in two human EBV-containing cell lines, either by transfecting ZEBRA expression vector into HKB5/Cl8 cells or HH514-16 cells (Fig. 1, A and C) or by treating HH514-16 cells with two chemicals that induce ZEBRA expression, namely sodium butyrate and phorbol ester (TPA) (Fig. 1B). The cells were labeled with [ 32 P]orthophosphate and then radioactively labeled ZEBRA protein was purified by immunoprecipitation followed by SDS-PAGE and excision from the gel. Two radioactive spots that co-migrated with the phosphoserine and the phosphothreonine standards were observed in both cell lines (Fig. 1). No radioactive spot co-migrated with the phosphotyrosine standard. These experiments showed that in two different cell lines, ZEBRA expressed from a transfected plasmid as well as ZEBRA expressed from the endogenous viral genome were phosphorylated at both serine and threonine residues.
In HKB5/Cl8 cells transfected with a ZEBRA expression vector the stoichiometry of phosphoserine and phosphothreonine was ϳ1 (Fig.  1A). In HH514-16 cells treated with chemical inducing agents, TPA and sodium butyrate, phosphoserine was more abundant than phosphothreonine ( Fig. 1B). Previous experiments have shown that TPA treatment leads to phosphorylation of Ser 186 (39). In HH514-16 cells transfected with an expression vector, phosphoserine was also more abundant than phosphothreonine in the ZEBRA protein (Fig. 1C). In all subsequent experiments we examined the constitutive phosphorylation state of ZEBRA that was expressed after transfection of HH514-16 cells in the absence of any chemical treatment that would induce the viral lytic cycle. Moreover, for these experiments we used the ZEBRA S186A mutant in which the major protein kinase C site is abrogated.
Strategy for Mapping Constitutive Phosphorylation Sites on ZEBRA Protein in Vivo-ZEBRA contains 15 serines and 12 threonines (Table  1). We used a strategy to identify in vivo phosphorylation sites on ZEBRA that did not require the necessity of mutating every serine and threonine individually and in combination. This strategy consisted of four interrelated maneuvers directed at identifying phosphorylated tryptic peptides. 1) We identified tryptic peptides of ZEBRA that, when labeled with [ 35 S]cysteine or methionine or with [ 32 P]orthophosphate, migrated similarly on a two-dimensional TLC plate (e.g. Fig. 2). These tryptic peptides presumably contain both cysteine (Cys) or methionine (Met), and serine (Ser) or threonine (Thr). The definitive identification of these tryptic peptides involved site-directed mutagenesis of Cys or Met and Ser or Thr associated with the disappearance of a radiolabeled peptide. 2) We identified 35 S-labeled NH 2 -terminal tryptic peptides by apposition of a 10-amino acid NH 2 -terminal FLAG tag (Fig. 3). The co-migration strategy allowed us to predict the identity of NH 2 -terminal phosphopeptides (e.g. Fig. 3). 3) Another component of the strategy, to differentiate between complete versus partial tryptic digestion products, involved mutating tryptic cleavage sites (e.g. Fig. 4). 4) Finally, we confirmed the identity of phospholabeled peptides suggested by the co-migration strategy by specific site mutations of serines and threonines (Fig. 5).
The strategy of 32 P/ 35 S peptide co-migration was facilitated if phosphopeptides either contain Cys or Met or are adjacent to peptides containing Cys or Met. Table 1 shows that four tryptic peptides (numbers 1, 3, 5, and 17) contain both Cys or Met and Ser or Thr and are therefore candidates for identification of phosphorylation sites by co-migration of 32 P and 35 S peptides. Five additional peptides (numbers 2, 4, 8, 14, and 18) contain Ser or Thr but lack sulfur-containing amino acids. These potential phosphopeptides could be identified using 35 S labeling if the tryptic peptides abut a Cys or Met containing peptide and are partial digestion products. Peptides 2, 4, and 18 fulfill this criterion. Peptide 4 is not constitutively phosphorylated nor is peptide 8, based on previous experiments. The site-directed mutants, T159A in peptide 4 and S186A in peptide 8, that abolish the protein kinase C sites do not alter the constitutive phosphopeptide pattern of ZEBRA (40).

TABLE 1 Peptides predicted from complete digestion of ZEBRA by trypsin
This analysis was obtained from EXPASY. Methionine and cysteine residues are shown in bold and underlined. Serines and threonines are in bold and italics. Peptides were predicted from complete digestion of ZEBRA by trypsin. Single amino acids were not assigned numbers.

Peptide
No. We studied the phosphorylation of ZEBRA in HH514-16 cells, an EBV positive Burkitt lymphoma-derived cell line, because the presence of EBV and cell background affects the phosphorylation status of ZEBRA (44). To prevent activation of ZEBRA protein from the latent EBV genome, we used a ZEBRA mutant, Z(S186A), which lacks the capacity to stimulate the induction of the lytic cycle and consequently is unable to activate the endogenous ZEBRA protein (45). The phosphopeptide patterns of wild type ZEBRA and the mutant Z(S186A) are identical (40). Therefore, the effect of additional specific point mutations was studied in the background of Z(S186A).
Comparison of Migration of 32 P-and 35 S-Labeled Tryptic Peptides of ZEBRA-Cells transfected with Z(S186A) were labeled either with [ 32 P]orthophosphate or with a combination of [ 35 S]methionine and [ 35 S]cysteine. The labeled proteins were subjected to immunoprecipitation with antibody to ZEBRA followed by tryptic peptide analyses. The 32 P-labeled tryptic peptide maps of Z(S186A) invariably contained five prominent groups of phosphopeptides labeled A through E ( Fig. 2A). Phosphopeptides B, C, and D were reproducibly more intensely labeled than phosphopeptides A and E. Further work will demonstrate that the spots encompassed by group E are related (see Fig. 5). The 35 S-labeled map of tryptic peptides of Z(S186A) contained approximately 19 peptides. The migration of both 32 P-and 35 S-labeled tryptic peptides could be arbitrarily divided into four vertical columns designated ␣, ␤, ␥, and ␦. By inspection, it can be seen that phosphopeptides B and E and 35 Slabeled peptides I through V are both located in column ␣. In experiments to be presented, we will identify 35 S-labeled peptides I-V by means of apposition of a FLAG tag and site-directed mutagenesis and we will explore the hypothesis that phosphopeptides B and E correspond to this group of 35 S-labeled peptides. By contrast, phosphopeptide C does not appear to co-migrate with any prominent 35 S-labeled peptide in column ␤, suggesting that phosphopeptide C lacks Cys or Met residues. We will identify phosphopeptide C by the strategy of abolition of a tryptic cleavage site.
Identification of Phosphopeptide D by Means of Co-migration with 35 S-Labeled Tryptic Peptide 5-By determining which tryptic peptides disappeared following site-directed mutagenesis of cysteine and methionine to alanine, we deduced the identity of three groups of 35 S-labeled tryptic peptides (Fig. 2B). A group of spots representing peptide 17 (P17) disappeared when Met 221 and Cys 222 were mutated to alanine; the spot corresponding to peptide 3 (P3) was eliminated in a 35 S-labeled tryptic peptide map analyzing a ZEBRA mutant with alanine substitutions at Cys 46 and Cys 132 (Table 1, data not shown). The 35 S tryptic peptides P3 and P17 (Fig. 2B) migrate in column ␥; however, no phosphopeptides were encountered in this column. Peptide 5 (P5), located at the base of column ␦ disappeared when Cys 171 was substituted with alanine. This result has been described previously (31). 35 S peptide 5 and phosphopeptide D are both in column ␦. Thus phosphopeptide D was identified by the co-migration strategy (31). 35 S-Labeled tryptic peptides that co-migrated with phosphopeptide spot A were unaffected when Cys 171 was mutated to alanine. When Cys 171 was mutated to methionine, which is more abundantly represented in the 35 S-labeled amino acid mixture, the intensity of the 35 S tryptic peptide that co-migrated with phosphopeptide spot D increased in intensity, but the 35 S-labeled peptides that co-migrate with phosphopeptide spot A were unaffected (see "Discussion").
Identification of NH 2 -terminal 35 S-Labeled Tryptic Peptides of ZEBRA-The NH 2 -terminal peptide of ZEBRA encompassing amino acids 1-12 contains two methionines, two serines, and one threonine (Table 1). Therefore, this peptide, if phosphorylated, is also a good candidate for identification by the co-migration strategy. To identify the location of the ZEBRA amino-terminal peptide, without mutating the first two methionines and hence jeopardizing the translation of the pro-tein, we appended a 10-amino acid NH 2 -terminal FLAG epitope tag. The FLAG tag has a methionine residue at position 1 and a lysine residue at position 4 (Fig. 3A). Therefore, the FLAG tag would be expected to have three major effects on the 35 S tryptic peptide pattern of the ZEBRA protein (Fig. 3A): (i) a new peptide corresponding to FLAG amino acids 1-4 would appear; (ii) a fusion peptide corresponding to FLAG amino acids 5-10 plus ZEBRA peptide 1 (P1) encompassing amino acids 1-12 would appear; (iii) the spots corresponding to peptide 1 of ZEBRA without a FLAG tag would be altered in mobility or disappear.
These changes can be seen by comparing Fig. 3, B, C, and D. Three additional prominent spots, designated F (1-4), FP1, and FP1Ј were found in the 35 S peptide maps of FLAG-Z(S186A) (Fig. 3C). We mutated Met 1 and Met 2 of ZEBRA to alanine in the background of FLAG-Z(S186A) to define which of these prominent new spots represented the first four amino acids of the FLAG sequences versus FLAG aa 5-10 fused to ZEBRA (Fig. 3D). Two spots, designated FP1 and FP1Ј, representing two forms of FLAG/ZEBRA chimeric proteins were eliminated; Complete digestion with trypsin would generate a 4-amino acid FLAG peptide, a chimeric 18-aa FLAG/ZEBRA peptide 1, and ZEBRA peptide 2 (amino acids 13-31). ZEBRA peptide 2 would not be visualized by 35 S labeling because it does not contain Met or Cys. However, a 37-aa chimeric FLAG/ZEBRA peptide containing sequences from ZEBRA peptides 1 and 2 could be visualized as a result of incomplete trypsin digestion at position 12. B-D, mapping of the positions of the tryptic peptide containing methionines 1 and 2 of ZEBRA. HH514-16 cells were transfected with Z(S186A) (B), FLAG-Z(S186A) (C), or FLAG-Z(S186A) with the first two methionines in ZEBRA mutated to alanine (D). The cells were metabolically labeled with 35 S, ZEBRA was immunoprecipitated and digested with trypsin. Comparing B and C note that peptides I-VI disappear in the FLAG-tagged versions. In panels C and D the first four amino acids in the FLAG tag are indicated by the peptide termed F1-4. FP1 and FP1Ј are the two chimeric peptides composed of the last six amino acids of the FLAG tag and the ZEBRA NH 2 -terminal peptide(s) containing methionines 1 and 2. FP1 and FP1Ј disappear following mutation of Met 1 and Met 2 to alanine (D). Bold dashed circles indicate the position of 35 S-labeled peptides that contain FLAG tag sequences.
the spot (F1-4) consisting of FLAG amino acids 1-4 remained. Six 35 S-labeled spots of Z(S186A) designated I-VI, present in column ␣ and the base of column ␤ in Fig. 4B, disappeared. Therefore, in the untagged version of the protein these six spots represent various forms of the NH 2 -terminal peptide 1 of ZEBRA. Fig. 3B) that disappeared when a FLAG tag was added to the NH 2 terminus of ZEBRA could represent either complete or partial digestion by trypsin. Whereas peptide 1 contains two methionines, pep- . Analysis of the NH 2 terminus region of ZEBRA by incomplete trypsin digestion. A, illustration of predicted NH 2 -terminal peptides generated by trypsin digestion and location of serines and threonines within these peptides. Depending on the extent of trypsin cleavage of wild type ZEBRA protein at lysine 12, two peptides will be observed with 35 S labeling, peptide 1 (I), or, peptide 1 ϩ 2 (II), that represents lack of digestion with trypsin at Lys 12 . Peptide 2 cannot be detected with 35 S because of the absence of methionine and cysteine residues from its amino acid sequence. In the mutant Z(K12A) (III), peptide 1 is absent and peptide 1 ϩ 2 is detected. B and C, 35 S-labeled tryptic peptide maps of wild type Z(S186A) (B) and a mutant ZEBRA, Z(K12A/S186A), in which lysine 12 was mutated to alanine in the background of S186A (C). Peptide 1 (P1) and peptide 1 ϩ 2 are indicated by dashed boxes. tide 2 lacks methionines or cysteines (Table 1). However, peptide 2 might be visualized by 35 S labeling as a result of incomplete trypsin digestion by virtue of its juxtaposition to peptide 1. To determine whether spots I-VI were the result of complete or partial digestion, the trypsin cleavage site, Lys 12 , was substituted with alanine (Fig. 4A). The 35 S-labeled tryptic peptides of Z(S186A) (Fig. 4B) were compared with those of Z(K12A/S186A) (Fig. 4C). This mutation resulted in the disappearance of a group of 35 S peptides located in the bottom half of column ␤ (Fig. 4C). Therefore, the group of peptides in the bottom of column ␤ can be assumed to represent various forms of the complete trypsin digestion product of peptide 1. Following abolition of the Lys 12 tryptic cleavage site the peptides in column ␣ remained; in fact, their intensity increased. Therefore the 35 S-labeled peptides in column ␣ represent a fusion of peptides 1 and 2.

Identification of 35 S Peptides Representing Incomplete Digestion of Peptide 1 Plus 2 by Trypsin-The group of 35 S-labeled tryptic peptides (I-VI in
Identification of Phosphopeptides Resulting from Phosphorylation in the NH 2 -terminal Tryptic Peptides of ZEBRA-The foregoing analysis predicted that prominent phosphopeptide B located in column ␣ was the result of phosphorylation in peptide 2. To test this hypothesis the pattern of phosphopeptides of the mutant Z(T14A/T30A/S186A) containing point changes in the two phosphoacceptor residues in peptide 2 ( Fig. 5B) was compared with that of wild type ZEBRA (Fig. 5A). Mutation of the two potential phosphorylation sites in peptide 2, namely Thr 14 and Thr 30 , resulted in the disappearance of phosphopeptides B and C (Fig. 5B). Phosphopeptide B represents the fusion of tryptic peptides 1 and 2; phosphopeptide C represents a complete digestion of tryptic peptide 2. A 35 S-labeled tryptic peptide corresponding to phosphopeptide C was not visualized (Fig. 2) as peptide 2 lacks cysteine and methionine residues.
To determine whether Thr 14 or Thr 30 were the phosphorylation sites, we studied a mutant in which Thr 30 was substituted with alanine ( Fig.  5C). This mutant, Z(S6A/T7A/S8A/T30A/S186A), contains changes in all the potential phosphorylation sites in peptide 1, as well as one of the two sites in peptide 2. This change did not affect the intensity of phosphopeptides B or C. Thus, by elimination Thr 14 is the phosphorylation site in peptide 2. However, the changes resulted in the disappearance of a group of phosphopeptides in the lower half of column ␣ designated E. The E phosphopeptides are likely to be the result of phosphorylation of Ser 6 , Thr 7 , and Ser 8 .
From the foregoing analysis we concluded that ZEBRA is constitutively phosphorylated in vivo on three tryptic phosphopeptides, peptide FIGURE 5. Mapping phosphorylation sites in the NH 2 terminus of ZEBRA. A-C, phosphopeptide maps of wild type ZEBRA, Z(T14A/T30A/S186A), and Z(S6A/T7A/S8A/T30A/S186A). The ZEBRA proteins were expressed in HH514-16 cells and labeled with [ 32 P]orthophosphate. Bold dashed circles denote the position of phosphopeptides that disappeared as a result of site-directed mutagenesis. D, illustration of ZEBRA phosphopeptides identified by this strategy. Numbers in boxes below the diagram represent the tryptic peptide number (see Table 1); letters in circles above the diagram represent the tryptic phosphopeptides (see Fig. 2) 1 (Ser 6 -Thr 7 -Ser 8 ), peptide 2 (Thr 14 ), and peptide 5 (Ser 167 -Ser 173 ) (Fig.  5D). These phosphorylations correspond to phosphopeptides E, B and C, and D, respectively.
JNK Reproduces ZEBRA Phosphorylation at Thr 14 in Vitro-Inspection of the amino acid sequence of ZEBRA contiguous to Thr 14 revealed the presence of a short consensus sequence, (Ser/Thr-Pro-Asp/Glu), similar to that of the c-Jun NH 2 -terminal phosphorylation sites, serine 63 and serine 73, both of which are subject to phosphorylation by JNK (46).
To test whether ZEBRA can be phosphorylated by JNK we carried out in vitro phosphorylation experiments (Fig. 6, A and B) with recombinant ZEBRA protein that had been expressed in E. coli and purified to homogeneity using nickel affinity chromatography. 25-100 ng of recombinant JNK efficiently phosphorylated 500 ng of the ZEBRA protein in vitro. Furthermore, 25 ng of recombinant JNK efficiently phosphorylated 50 -250 ng of ZEBRA (data not shown). When recombinant wild type and the T14A mutant ZEBRA proteins were compared, the overall level of phosphorylation of the mutant by JNK was reduced by ϳ60% (Fig. 6, C and D). These results suggested that Thr 14 serves as a substrate for in vitro phosphorylation by JNK. To confirm and extend this result, the phosphopeptide patterns of wild type and T14A mutant ZEBRA proteins were compared after in vitro phosphorylation by JNK. Threonine to alanine substitution at position 14 resulted in the disappearance of 4 phosphopeptides, numbered 1 to 4. In wild type, ZEBRA phos-phopeptides 3 and 4 were more intense than phosphopeptides 1 and 2; their migration was similar to phosphopeptides B and C observed in phosphopeptide maps of metabolically labeled ZEBRA from eukaryotic cells. This result suggests that in vivo phosphorylation of ZEBRA at Thr 14 can be recapitulated by incubating the protein with JNK in vitro.

DISCUSSION
ZEBRA, a transcription activator and replication protein encoded by Epstein-Barr virus, is constitutively phosphorylated in vivo. Only one of the constitutive in vivo phosphorylation sites has previously been mapped to a single phosphopeptide containing Ser 167 and Ser 173 (31). These residues located in a regulatory domain of ZEBRA can be phosphorylated in vitro by CK2 (47). Phosphorylation at the CK2 site is essential for the ability of the ZEBRA protein to serve as a repressor of late genes activated by the other EBV immediate early protein, Rta (31). Our recent unpublished data 3 indicates that phosphorylation at CK2 sites also influences temporal regulation of other components of the lytic cascade, such as lytic replication and late gene expression. Phosphopeptide maps of metabolically labeled ZEBRA revealed the presence of multiple unidentified phosphorylation sites whose position in the protein and the responsible kinase had not been characterized. In this  and a previous report (31), we have identified the constitutive phosphorylation sites of ZEBRA. We found that ZEBRA is phosphorylated in vivo at serine and threonine residues clustered in two domains of the protein, the transactivation domain and the regulatory domain. Phosphorylation in the regulatory domain is restricted to the two previously characterized phosphorylation sites, Ser 167 and Ser 173 (31). The transactivation domain harbors a cluster of minor phosphorylation sites (Ser 6 -Thr 7 -Ser 8 ), and a major phosphorylation site (Thr 14 ). The Thr 14 site contains a carboxyl-terminal sequence similar to the JNK phosphorylation motif in the transactivation domain of c-Jun, namely (Ser/Thr-Pro-Asp/Glu). In vitro phosphorylation experiments confirmed that Thr 14 of ZEBRA may serve as a substrate for phosphorylation by the stress-activated protein kinase, JNK.
Strategy Employed to Identify in Vivo Phosphorylation Sites-In this report we followed a co-migration strategy to map in vivo phosphorylation sites of a protein that was not abundantly expressed or purified. An essential component of this strategy was to mutate cysteine and methionine residues in each peptide, thus enabling their identification in 35 Slabeled tryptic peptide maps. The migration of 35 S-labeled peptides was then compared with that of unknown phosphopeptides. Co-migration provided useful clues to the identity of the unknown phosphopeptides. For example, the identity of phosphopeptide D was predicted by identifying the co-migrating 35 S-labeled peptide as peptide 5 (31). Similarly, the identity of phosphopeptides B and E was predicted by virtue of their co-migration with 35 S-labeled peptides representing the NH 2 -terminal region of ZEBRA in column ␣ (Figs. 4 and 5). This strategy overcame the requirement for mutating each phosphoacceptor residue as it focused on the identification of specific peptides rather than individual residues. Unlike phosphorylation, 35 S incorporation takes place during the synthesis of polypeptides and thus is independent of conformational changes of the target protein. Therefore, the co-migration approach can distinguish between elimination of a primary phosphorylation site and disruption of a phosphorylation event induced by a conformational change in the protein.
Despite considerable information provided by this approach, a few limitations exist. The distribution of methionine and cysteine residues relative to serines, threonines, and tyrosines in the protein of interest was important. A peptide that contains a potential phosphorylation site but lacks sulfur containing amino acids cannot be detected by 35 S labeling on a TLC plate. For example, peptide 2 of ZEBRA does not contain methionines or cysteines and therefore could not be detected by 35 S labeling (Fig. 4). However, incomplete trypsin digestion facilitated the identification of this peptide through the formation of a fusion peptide composed of peptide 2 and 35 S-labeled peptide 1.
Phosphopeptide A-Occasionally mutation of phosphorylation sites in a single peptide can result in the disappearance of two or more phosphopeptides. This might occur as the result of elimination of a primary phosphorylation site that is essential for the phosphorylation of the protein at a secondary site. The second phosphopeptide may represent alternative peptide forms that result from additional post-translational modifications that are contingent on the primary phosphorylation site.
This situation was encountered with mutations in the CK2 sites in peptide 5. Ser 167 and Ser 173 are inseparable by trypsin and present on the same peptide; yet mutations at Ser 167 and Ser 173 to alanine abolished two phosphopeptides, A and D. Whereas phosphopeptide D co-migrated with a 35 S-labeled peptide, as indicated by the C171A mutation, no additional 35 S-labeled peptide corresponding in mobility to phosphopeptide A was detected. These results suggest that spot A represents a phosphopeptide whose phosphorylation is dependent on primary phosphorylation of the CK2 site. Because spot A did not correspond to a 35 S-labeled peptide it may not contain or be adjacent to peptides containing methionine or cysteine.
Thr 14 of ZEBRA Is a Potential JNK or Other MAP Kinase Site-We found that threonine 14 is a major constitutive phosphorylation site of ZEBRA, as indicated by the intensity of phosphopeptides B and C. The presence of a proline residue carboxyl-terminal to Thr 14 suggests that this site might constitute a minimal consensus sequence for phosphorylation by MAPKs (proline-directed kinases) (48). Threonine 14 in ZEBRA, and serine 63/serine 73 in c-Jun, share the same minimum MAPK consensus sequence, namely, Ser/Thr-Pro-Asp/Glu. Our in vitro phosphorylation experiments demonstrated that JNK efficiently phosphorylates ZEBRA at Thr 14 . However, unlike c-Jun, ZEBRA lacks a ␦-domain, an independent sequence located upstream of the phosphorylation site that is thought to serve as a docking site for JNK. The ␦-domain mediates the interaction between substrate proteins and JNK; its presence is required for efficient phosphorylation by JNK (49 -52). This result provokes the idea that ZEBRA might represent a new class of JNK-substrate proteins that contain a non-canonical docking site. Extracellular signal-regulated kinase, another member of the MAPK family of protein kinases, has been shown to employ multiple motifs as docking sites to interact with different substrates (52,53). Therefore, understanding the mechanism by which ZEBRA interacts with JNK might lead to the identification of new substrates for JNK.
Despite the absence of any significant homology between the transactivation domain of ZEBRA and c-Jun, both domains may now be considered to share two important features: interaction with CBP and phosphorylation by JNK (3,46). In c-Jun, both of these features are interrelated because phosphorylation of c-Jun by JNK at Ser 63 and Ser 73 enhances the interaction of c-Jun with CBP and thus augments the transcriptional activation function of the protein (3). Similarly, the interaction between ZEBRA and CBP enhances the transcriptional activity of ZEBRA by recruiting the intrinsic histone acetyltransferases activity of CBP to ZEBRA responsive promoters (54 -56). However, additional experiments are needed to determine whether phosphoryla-  FEBRUARY 10, 2006 • VOLUME 281 • NUMBER 6 tion of ZEBRA at Thr 14 by JNK regulates its interactions with CBP or other histone acetyltransferases. More work is also required to prove that Thr 14 is a bona fide JNK site in vivo or whether Thr 14 can serve as a phosphoacceptor for another member of the MAPK family. Phosphorylation of c-Jun by JNK at Ser 63 and Ser 73 has been implicated in several other aspects of the behavior of the protein, such as protein stability and inhibition of SUMO-1 conjugation (57,58). A c-Jun mutant that cannot be modified by sumoylation at Lys 229 exhibited enhanced transcriptional activation function (57). In a process analogous to c-Jun, JNK-mediated phosphorylation of ZEBRA at Thr 14 might influence the half-life of ZEBRA or block its modification by SUMO-1. A previous report demonstrated that ZEBRA is modified by sumoylation at Lys 12 (59). One hypothesis is that SUMO-1 conjugation might serve as a mechanism by which the cell turns off the transcriptional activity of these proteins. However, phosphorylation of c-Jun and ZEBRA by JNK at Ser 63 -Ser 73 or Thr 14 , respectively, might serve as a defense mechanism to block SUMO-1 conjugation and thus maintain the transcriptional potential of these two proteins. Considerable additional work will be required to unravel the functional consequences of the phosphorylation sites that have been identified in this study. In particular it will be important to explore the significance of the Thr 14 site on the many functions of the ZEBRA protein that have been described, including activation and repression of transcription (29,31,60,61), activation of DNA replication (25,62), regulation of cell cycle (63)(64)(65), and activation of cellular kinases and immune modulation (66,67).

ZEBRA Constitutive Phosphorylation Sites
In summary, ZEBRA phosphorylation can be divided into two main categories, constitutive and inducible. Constitutive phosphorylation of ZEBRA transfected into HH514-16 cells results in five tryptic phosphopeptides, A-E. Phosphopeptide A has not been characterized yet; phosphopeptides B and C correspond to phosphorylation at Thr 14 ; phosphopeptide D, Ser 167 -Ser 173 ; and phosphopeptide E, Ser 6 -Thr 7 -Ser 8 (Fig. 5D and Table 2) (31). However, it remains possible that phosphopeptide E is phosphorylated at threonine 14 as well. The stoichiometry of phosphorylation of ZEBRA on these three tryptic peptides is an important subject for further studies. Treating ZEBRA-expressing 293 cells with phorbol esters, a protein kinase C agonist, induces ZEBRA phosphorylation at Ser 186 and results in the appearance of two additional phosphopeptides ( Table 2) (39). In vitro phosphorylation experiments revealed that threonine 14 and serines 167/173 are substrates for phosphorylation by JNK and CK2, respectively ( Fig. 6) (31,47), whereas Thr 159 and Ser 186 are subject to phosphorylation by protein kinase C (39,40). Thus the many functions of ZEBRA are likely to be regulated by constitutive and inducible phosphorylation.